OpenRouter

A month of gateway changes in one place: audio endpoints, parallel multi-model Model Fusion, private fine-tuned endpoints, a router that picks the cheapest model clearing a coding score, enforced IP allowlists, and per-provider zero data retention toggles.

Ox Alpha is an anonymous frontier model on OpenRouter aimed at efficient coding and sustained agentic work, with a 1M token context and text, image and video input. It is free during the stealth window, and the provider does not train on prompts or completions.

Descript's Underlord video agent went from three hardcoded model integrations to 13 production models across four labs by routing inference through OpenRouter, cutting evaluation time from a week to under two hours.

OpenRouter published Model Router Benchmarks comparing seven routers, including NVIDIA Switchyard and Unbiased Pareto, on quality, speed and cost across six benchmarks, plus model selection and reliability views.

OpenAI's Decisions API, now in public beta and listed on OpenRouter as GPT-6 Luna Decisions, returns typed decisions with probabilities for text, JSON or image inputs. Claimed up to 10x faster than GPT-6 Luna via Responses, at $0.10/M input with free output.

OpenRouter added a /tools API that lets agents pull Server Tools Marketplace data, including best-in-market prices, and use it programmatically. Docs and a browsable marketplace are linked.

OpenRouter's US/EU In-Region Routing is now GA: requests are decrypted and processed entirely inside the chosen region from TLS termination through the provider. Switch via base URL; unsupported models 404 instead of leaving the region. Enforceable per workspace via Guardrails.

OpenRouter ran an identical prompt through 20 image generation models and read the real billed cost off each response, finding a 22x price spread and naming picks for cheapest, best editing, and text-in-image output.

OpenRouter walks through building a marketplace moderation system with the Jev API: structuring state, writing questions that survive literal reading, calling the TypeScript API, and calibrating publish/hold/reject thresholds from labeled data.

Five human-in-the-loop patterns for agents facing EU AI Act Article 14, Colorado SB26-189 and the NIST AI RMF: classify tools by risk tier, log every oversight event, escalate on timeout, back approval state with durable storage, and wire the gate into the Agent SDK loop.

OpenRouter's December roundup leads with Claude Code support for usage tracking and cost control, plus automatic JSON repair for structured outputs, browser notifications on long completions, and rankings filtered by context length.

PDF input works across all OpenRouter models, falling through to native handling for providers that support it and OCR for the rest. Two engines set the cost floor: mistral-ocr at $2 per 1,000 pages, or free text-only extraction with no image support.

The largest multi-model gateway is joining Stripe, with no announced change to the API, product or roadmap. Relevant mainly as dependency risk: routing neutrality across 400+ models is now a commitment made by a subsidiary rather than an independent company.

A five-step method for routing between a cheap and a strong model using a structured-output confidence field, with thresholds derived from your own error rates and re-tuned in production.

OpenRouter labels which models allow their outputs to be reused as training data and adds enforce_distillable_text to hold a request to that set. The post pairs it with NVIDIA NeMo Data Designer for generating structured synthetic datasets and distilling large models into task-specific ones.

OpenRouter benchmarks TypeSafe's Jev decision model against Claude Opus 5 on 3,080 Banking77 classification utterances, finding Jev trails by 3.3 accuracy points but runs 13x faster at 1/22 the cost.

OpenRouter reports Anthropic's Opus 5.5 leads all models in share of platform spend, with Sonnet 5.5 and Haiku 5.5 also launching strongly, pointing to a rebound against OpenAI in real usage.

OpenRouter's January changelog: partition sorting deprioritises providers that miss a throughput or latency threshold, the auto router takes wildcard model filters across 58 models including Opus 4.5, the SDK gains an idempotent skills loader, and p99 gateway latency drops sharply.

Reasoning traces become a first-class field on OpenRouter via include_reasoning, starting with DeepSeek R1 and its derivatives. Sample code shows harvesting R1's reasoning and injecting it as context for a much weaker model to lift its answers.

Deep research becomes an API call rather than a product surface: Perplexity's Sonar Deep Research, Reasoning Pro and Pro arrive on OpenRouter with search and reasoning steps priced separately, joined by Cohere Command A, Jamba 1.6 and free Gemma 3 models.

The @openrouter/agent SDK collapses the agent loop into one callModel function: Zod-validated tools, validation of arguments the model produces, composable stop conditions such as stepCountIs and maxCost, concurrent text, tool-call and reasoning streams, and cost data per run.

OpenRouter adds retrieval to every model through an :online suffix or web plugin, retrieving via Exa at $4 per 1,000 results. The injected system message is published and configurable, along with the result count, so you can control cost and prompt shape.

OpenRouter details NVIDIA's Nemotron 3.5 Lightning, a 30B mixture-of-experts model with 3B active parameters aimed at cheap, high-volume agent steps like tool calls and validation, contrasting it with the heavier Nemotron 3 Ultra.

OpenRouter makes provider selection controllable per request: :nitro sorts providers by tokens per second, :floor by price, and provider.sort exposes price, throughput or latency directly. Choosing a sort turns off the default load balancing.